SYNCHRONIC vARIATION: SYDNEY BASIN ENGRAvED ARTpress-files.anu.edu.au/downloads/press/p109931/pdf/ch111.pdf · terra australis 27 253 Chapter 11: Synchronic variation: Sydney Basin

terra australis 27

251

Chapter 11: Synchronic variation: Sydney Basin

11 SYNCHRONIC vARIATION: SYDNEY BASIN ENGRAvED

ART

IntroductionThis chapter looks at regional stylistic variability in the engraving component. The investigation is restricted to motif depictions, given that technical variation here is minimal. Analysis was aimed at identifying whether broad scale patterns could be interpreted culturally, not just environmentally (McMah 1965). A comparison of these results with those achieved in the shelter art assemblage is made later (Chapter 12). There was initial concern that focussing on motif would not investigate style per se, but mere compositional variety33. Given the overall aim was to analyse the engraving and shelter art components at a comparable level (despite their technical differences and the variability that these introduce to motif form), motif taxonomy appeared the most judicious approach to the problem. Motif has been successfully employed at a regional and localised level, investigating a range of stylistic questions (Clegg 1987, Officer 1984, Franklin 1984, Smith 1989). Sackett(1990) cites various examples of how the combination of motifs and compositional features may indicatehighlevelsofethnicallysignificantpatterning(e.g.Glassie1975,Longacre1981)andsuggests that ‘themes may well be the [things] that give congruence to isochrestic choices in non-materialaspectsofculturallife’(Sackett1990:41).Whilethistypeofclassificationhadnotbeenpreviously attempted on an archaeological rock art assemblage, the approach used here tested Sackett’s proposition. Correspondence Analysis (CA), which seeks patterning in the combination of variables (motifs) in the data set, was considered an ideal tool for this analysis. The CA indicated that the region’s engraving assemblage was largely homogeneous (see below). The results were viewed according to the language boundaries in evidence at white contact and to major drainage basins within the study area. These contexts provide an explanatory framework for variation across the region. The posited style boundary to the south of the region was further investigated. This was initially identifiedbyMcMah(1965).TheRockArtProjectdetermined the likely locationforthis boundary was the Georges River - and that this boundary existed for both art components (McDonald 1985a, 1990a).This research refines further the extent of the stylistic differencesoneithersideofthisboundary.Styleclinesarealsodefinedanddescribedelsewhereacrosstheregion. These are manifested as increasing and/or decreasing amounts of homogeneity in localised areas and varying motif foci. Engraving sites in different topographic locations were investigated to explore the possibility of different social contexts for information exchange. Vertical engraving sites around the estuarine foreshores (in very public locations) were compared with open engraving sites on the ridgelines (where it is assumed that the audience was more restricted). The distribution of uncommon motifs is also explored. Rare motifs were thought to have the bestpotentialtodemonstratetheinfluenceoflocal(orevenindividual)stylistictraits.Raremotifs33A qualitative approach to individual motifs (e.g. body proportions, angle of macropod’s tails, orientation of motif and presence of eyes and other internal features, etc.) may also have revealed stylistic patterning. Given the lack of success of this approach to demonstrating ethnicity on a limited scale (Smith 1983; and see Clegg 1981), and the generalisedoutcomeachievedbysuchanapproachataregionallevel(Franklin1984),thisadditionaltypeofanalysiswas not attempted.

terra australis 27

252

Dreamtime Superhighway: an analysis of Sydney Basin rock art and prehistoric information exchange

are swamped in large-scale analyses, and it was hoped that this type of analysis would provide additional stylistic information. Whales, for instance demonstrate a fairly restricted geographic range. This, however, could be an environmental range as much as a cultural one. Other motifs, such as profile people, culture heroes, certain material objects and Complex-non-figuratives(CXNF’s)wereconsideredtobebettergaugesofculturalchoices.Therestricteddistributionofcontact motifs is also discussed. As well as the regionally based quantitative analyses, several geographically restricted qualitative analyses of motif depiction and preference (Sackett’s ‘compositional features’) were undertaken.Certainattributesonhumanfigures(particularlygenderanditemsofmaterialculture),CXNF’sandcultureheroesarethefocusofthisanalysis.

Defining a regional style: methodologyThe aim of these analyses was to provide a statistical description of each art component (i.e. average assemblage size, motif frequencies etc.) and to determine the amount of variation within the assemblages on the basis of multivariate analysis. The general approach described here was used for both art components. Aswithmostexercisesofquantification,logicalstepsarerequiredtocodethedatasoitcan be read by a computer, analysed and then interpreted. As well as logically ordering the data, it is necessary to justify the selection of variables as meaningful and relevant to the questions beingasked.Theclassificationandselectionprocesswasdiscussedfullyintheoriginalresearch(McDonald 1994a: Appendix 1). The procedures followed in quantifying the two art components canbebroadlydefinedas:

identificationofthesample;selection of variables;collection (counting) of variables for the sample;input of counted information into a (computerised) data base;selection/clumping of variables for analysis;analysis of data base; and,interpretation of analysis results.

Having identified the sample (Chapter5), the selectionofvariables foranalysisproceeded.Amotifclassificationwasappliedtobothartcomponents(thesewerefundamentallythesamewithaddition of several motifs for the more diverse shelter art component) and a taxonomy accounting for technical variations for the shelter art sites (see McDonald 1994a: Appendix 1). Topographic, grid reference and site association information for both site types was also collected (see McDonald 1994a: Appendices 5 and 6).

Motif Variables

While comparability between the two components was an overriding factor in the selection of motif variables, it was recognised that inherent differences in the two components would necessitate somevariationinthemotifsidentifiedandcounted. ThemotifclassificationwasinitiallydevisedfortheRockArtProject,specificallyfortheengravedassemblage(McDonald1985a).Thefieldrecordingexercisesundertakeninsubsequentstages of the Project revealed greater motif variability, particularly in the shelter art component (McDonald 1987, 1990a) which necessitated two additional motif categories (hands and axes) for both components, and the addition of two exclusive categories (hand stencil variations and ‘other’) for the shelter art sample (see Tables 5.1 and 5.3, Chapter 5).

terra australis 27

253


Counting

Oncethemotifandtechniqueclassificationshadbeendevised,countingproceeded34. The counted data was initially recorded on accounting broadsheets. Each site had one column; variables were recorded by row. As well as counting motif and technique variables, site card information (for those sites not visited by me) provided the topographic and site association variables. AHIMS (thenNPWS)siteidentificationwasusedforeachsite. All data were analysed on the mainframe computer at the Australian National University.

Analyses

Analysis commenced with the motif count information, i.e. the raw data which had been entered into the computer. This involved the analysis of 27 engraved motif variables. All data were put through a GENSTAT Correspondence Analysis (Version 3.1), designed to investigate variance within large multivariate data populations (the multivariate analyses used weredefinedanddiscussedinMcDonald1994a:Appendix1).Thismethodfoundthatbothartcomponents were largely homogeneous populations, which clustered tightly and showed no underlying structure. The pattern of distribution, both for variables and sites, was affected largely by the presence of a few outliers. The analysis was re-run removing the outliers, in an effort to seek the underlying patterns within the greater data set. This was done a few times, until it became clear that the method was imposing structure upon the data - in much the same way as a cluster analysis. Two further steps were taken in the treatment of the data. This involved clumping the motifs and converting them to binary data. The motif taxonomies for both components were reduced to seven variables (Table 11.1).

Converting to binary data was a simple process of using presence and absence rather than raw count data: if a variable was present it received a value of 1; if it was absent its value was 0. The CA of the engraving sites was based on a reduced sample of 705 sites (i.e.siteswithonlyunidentifiedmotifswereexcluded from the analysis).

Correspondence Analysis (CA): data, results and interpretationThe aim of a CA was to investigate sources of variance within the data set, to identify groups of similar and dissimilar objects (i.e. sites). The advantage of this technique over

other multivariate tools is that the variables (i.e. motifs) which contribute to these groupings can beidentified.Itisnotsomuchthepresenceofindividualmotifswhichcreatesthevariance,butthe combination of variables (Baxter 1994, Benzecri 1992, Shennan 1988). To explore the geographic variabilitywithin this assemblage, first aCAwas run usinggrid location as a factor. This revealed no internal groupings or structure, nor any strong positive correlation between geography and motif. This result supported the absence of strong patterning or major divisions within the assemblage.

34All sites including those which had been counted previously (McDonald 1985a) were recounted for this research.

Variable No. Motif/Variable description1 Anthropomorphic2 Terrestrial3 Birds4 Marine5 Material Objects6 Tracks7 Other

Table 11.1: engraving Sites. clumped motif variables used in correspondence Analysis.

Clumped variable 1 includes individual variables 1 - 5; 2 = v 6-8; 3 = v 9,10; 4 = v 11-14; 5 = v 15-8; 6 = v 21-24; 7 = v 25-27 (see Table 5.1). Unidentifiable motifs have been excluded from this level of analysis.

terra australis 27

254


The data base for this component is too large for the plot of the site distribution to be meaningfullyinterpreted(Figure11.1).Theresultswerethusreplottedusingsubdivisionsofthedata(Table11.2:seeMcDonald1994a:Appendix7;FiguresA7.1-A7.7).Thesesub-plotsarebasedon exactly the same results, but the smaller sample sizes enabled more useful interpretation. Analysiswasundertakenonavarietyof scales.Thefirst analysis involvedanarbitrarydivision of the sites into map sheet provenance. The data were subdivided into eight groups which could be interpreted broadly on a geographic basis. The resultant groups were unequal in size, indicating e the geographic biases of the sample (Table 11.2). While being arbitrary in terms of archaeological context, this division of the data base gave control over north-south and east-west clinesinthedata(seeFigure11.2;McDonald1994a:Appendix8). A more detailed subdivision of the region’s data was contemplated, but given concerns about the relevance of the documented linguistic boundaries (Chapter 3) and obvious sampling issues (i.e. unequal sample sizes), division of the data into archaeologically meaningful zones was a vexed question. To investigate possible cultural divisions across the region more closely, several localised areas (with good sample sizes) were chosen to investigate linguistic boundaries.

Figure 11.1: CA Scores: entire Engraving component (705 sites). Each dot represents many sites.

Table 11.2: Analytical grouping of engraving sites according to AhImS numbers. These groups were used in the regional interpretation of the cA results.

Group Map numbers 1:250,000/1:100,000 maps Sample size

Group 1 37 - 6 -’s Singleton/Cessnock 4 sites

45 - 1 -’s Sydney/Wallerawang

Group 2 45 - 2 -’s Sydney/St Albans 37 sites

Group 3 45 - 3 -’s Sydney/Gosford 234 sites

Group 4 45 - 4 -’s Sydney/Blue Mountains 10 sites

45 - 5 -’s Sydney/Windsor

Group 5 45 - 6 -’s Sydney/ Sydney 377 sites

Group 6 52 - 2 -’s Wollongong/Wollongong 19 sites

Group 7 52 - 3 -’s Wollongong/Port Hacking 35 sites

terra australis 27

255


Regional Analysis

Thefirsttwocomponentsaccountfor64%ofthevarianceinthesampleandthesecomponentsdiscriminatewell.Thefirstcomponentaccountsforthegreatestamountofvarianceinthedatabase(Figure11.2),andthescreeslopeplot(Wright1992)demonstratesthatthiscomponentdescribesconsiderable structure in the data. Less variance is accounted for by the second component, and the slope then tails off.

00.10.20.30.40.50.60.70.80.9

1

1 2 3 4 5 6 7

Whilenomajorinternalgroupingswereidentified,certainsiteswereidentifiedasoutliers(Figure11.1).Theplotofthefirsttwoco-ordinates(Figure11.3)revealsthatthreeofthevariables(2,3and4)aregooddiscriminators.Inthefirstco-ordinate,marineanimals(3)andbirds(4)arenegatively correlated, while in the second co-ordinate, birds (3) are negatively correlated with land animals (2). Thus sites which contain large numbers of bird motifs would have very few marine depictions, while sites with a large number of land animals would contain small numbers ofbirddepictions(andviceversa).Anthropomorphsarerelativelyweaklypositionedonthefirstco-ordinate, but have a relatively good discriminating effect on the second co-ordinate. Material objects, tracks and ‘other’ motifs are poor discriminators being situated close to the origin. The CA results reveal no evidence for strong or distinctive localised variability across the region. Certain stylistic clines can be observed in the region’s subdivisions as shown by the bivariate sub-plots (McDonald 1994a: Appendix 7: A7.1-A7.7) in terms of the distribution and/orclusteringofsitesrelativetotheorigin(i.e.X+Y=0onthebivariateplot).Sitesclosetotheorigin are poorly discriminated by the axes in question and are stylistically homogenous. Those site distributed away from the origin are well discriminated by their motif assemblage i.e. are stylistically differently. ‘Common’ sites fall close to the origin; unique and/or more unusual sites arelocatedawayfromtheorigin.Theidentificationanddistributionofoutliersitesinlocalisedareas (according to quadrants on the graph) is the key to investigating thematic variety across the region. Thebivariateplots foreachgroupwereanalysed.Thenumberofsiteswithinadefinedand consistent radius of the origin was noted35, allowing for a calculation of the percentage of ‘common’ and outlier sites in each area. This was necessary for comparability given the disparate sample sizes. As the computer generated plots sometimes generated the two axes at different scales, the radius is sometimes described by an ellipse rather than a circle. The distribution of theoutliersitesaccordingtothefourquadrantswasalsoinvestigated(Figure11.4).Variationsinthedistributionofsiteswithinthequadrantsidentifieddifferentcompositionalfociacrosstheregion. The sites in the north-west of the region were found to be relatively heterogeneous. Outlier sites occurred predominantly (86%) on the positive side of the vertical axis, indicating that they aremorestrongly (andpositively)discriminatedby thefirstcomponent.Sites in thesegroupscontainedmanytracks(birdandmacropod)aswellasadefinitepreferenceformacropodsandother35This was drawn on each of the bivariate plots as a heuristic device.

figure 11.2: engraving cA results: plot of the latent roots indicating that the variance in the data set is well accounted for by the first two components.

terra australis 27

256


land animals, birds and anthropomorphs. The large group of sites (234) in the north-east of the Basin (north of the Hawkesbury River) were more heavily clustered around the origin (45%), while the outliers showed a strong tendency (76%) to be located on the positive side of the vertical axis. There was a preference in these sites for tracks (particularly mundoes), material objects (shields) and marine depictions. Anthropomorphic depictions were also very common. The 377 sites in the centre of the Basin (located between the Hawkesbury River and Botany Bay; the Cumberland Plain and the coast) probably represent the core Sydney Basin engraving assemblage. Most of the sites (54%) in this group were tightly clustered around the origin. The majority of the sites in this group contained marine depictions, material objects, tracks (particularly mundoes) and anthropomorphs. Land animals were quite common, as were birds. There are several outlier groups within this sample. One major cluster (53 sites) contains combinations of exclusively marine depictions. In another major cluster (of 25 sites) each contains a single macropod.

South of the Georges River, there was a marked increase in heterogeneity. South of Port Hacking (group 8) the sites are well dispersed away from the origin (only 26.5% homogeneity). These sites contain a predominance of marine depictions and other material objects. Some of these sites included anthropomorphs; others included terrestrial animals and ‘other’ motifs. The outlier sites in this group indicate a major difference from preceding groups in subject preference. Whenthepercentagesofhomogeneoussitesacrosstheregionarecompared(Figure11.5),thispatterning is clear. The most homogeneous assemblages are in the centre of the Sydney Basin (Groups 5 and 3). This homogeneity decreases as you moving north-west and south. Sites to the west of the central core are also fairly homogeneous. There was also marked variation in the distribution of outlier sites (compositional focus) across the region. The emphasis on certain combinations of motifs varies across the region, and it is the changes in these combinations that characterise the stylistic clines across the region. These are sometimes explicable in terms of economic/geographic factors (e.g. proximity to the sea). Not all variations, however, were so easily explained.

language Areas. Searching for boundaries and between-group distinctivenessFivelanguageareas(followingCapell1970)arerecognisedtohaveexistedwithinthestudyareaatEuropean contact (chapter 3). These languages were mutually intelligible, although ethnohistoric evidence suggests that the locations of neighbouring ‘tribal territories’ were recognised and respected by the various groups. Anthropological work elsewhere on the continent suggests that the boundaries between language groups would not have been impenetrable barriers. Such studies

figure 11.3: cA results: engraving Sites. Bivariate plot of variable Scores.

figure 11.4: Quadrant labels used in the following discussion of the cA results.

A B

CD

(-+) (++)

(+-)(--)

-0.03

-0.02

-0.01

0.01

0.02

0.03

0.04

0.05

-0.06 -0.04 -0.02 0 0.02 0.04

terrestrial

anthropomorph

birds

marine animals

material objectstracks

other

terra australis 27

257


alsosuggestthatsuchboundariesmayhavebeenfluidovertime.Theterritorialdistributionofthecontact languages may not have extended back more than a few generations.

figure 11.5: percentage of homogeneous engraving sites in each analytical Group.

Archaeological evidence also casts uncertainty onto the longevity of these boundaries: the cultural changewhichoccurred at around1,000yearsBP (with the introductionoffishhooks and thedecreaseintheuseofrockshelters)maybesosignificantthatthecontactlanguageboundariesare meaningless throughout the full extent of the region’s art’s production (i.e. the last 3-4,000 years). Capell’s (1970) language group boundaries were major rivers and creeklines. I argue (following Tindale 1974, Peterson 1976) that the boundary of any group’s range is more likely to be at the periphery of its economically viable area. Boundaries in topographically dissected areas (such as the Sydney Basin – as with the Pilbara) are likely to be along ridgelines. The ethno-historically reported use of ridgelines for access routes around the region supports this argument since it is likely these would have traversed the periphery of any particular group’s territory – not bisected the centre. The model proposed in this thesis suggested that stylistic behaviour which reinforces group distinctiveness should be observable on the basis of drainage basin catchments, with boundaries between groups along ridgelines and not creeklines. The region was thus subdivided by means of drainage basins. Within thefivedocumented languageareas,25drainagebasinsweredefinedacross theregion. Asidentifiedearlier,apotentialproblemwiththeseanalysesisthedisparatesamplesizes(Table 11.3). This distribution of sites represents in part an archaeological ‘reality’ (Chapter 5), withdecreasingsitenumbersattheperipheryoftheSydneyBasin.Thismayinpartreflectthegeologicalreality,althoughthishasneverbeenquantified.Muchofthebiaswiththiscomponentresults from the geographic focus of certain recorders in locations closer to Sydney city (McDonald 1985a).Over70%ofthisassemblagewasrecordedbyW.D.Campbell,FredMcCarthyandIanSimmeansthatthedistributionofthissamplelargelyreflectstheirareasofinterestandrecording

terra australis 27

258


focus [Sim (1966a) represents the only focus in the north-west of the Basin]. More recent recording work (e.g. Tacon et al. 2006) indicates that there are still many engraving sites to be found in the west of the region. Several systematic EIS surveys, in the north and south of the region have revealed a relatively low number and density of engraving sites. Likely explanations for this could be geological or cultural. In the Mill Creek valley (south of the Georges River), the

ridgelines on either side of the drainage basin are characterised by shale laterite, some of which had been extensively mined by the local Municipal Council for road construction (Attenbrow and Negerevich 1981; McDonald 1990b). Systematic survey here (covering c. 45 sq km) revealed only one engraving site. In the north of the region, the Mangrove Creek valley, both in its upper catchment and its middle reaches around Warre Warren Creek, has been systematically sampled over an area of approximately 200 sq km (Attenbrow 1981, McDonald 1988a). Engraving sites here were relatively few (n=4: 3% of recorded sites) although large numbers of suitable rock surfaces were located and inspected. These results are in stark contrast with those achieved at Maroota south of the Hawkesbury River in the central-west of the Basin, where 12 engraving sites were located in one square kilometre (McDonald 1986a). Acorrelationofsamplesizewith languageareaandmotifcategorieswasmade(Figure11.6), to determine whether patterning in these analyses results from sampling inequities (James 1993).

Number of Motif categories recorded

Sam

ple

size

(No

of s

iites

)

050

100150200250300350400450

0 10 20 30

Thisdemonstratednodirectcorrelationbetweensitesizeandmotifnumbers.Further,theGuringai sites were subdivided into random 50 and 100 site samples, for the comparison of assemblage composition (and CA results) according to language area. These different steps indicate that the unavoidablesamplinginequitiesinthedatabasedonotappeartoproducesignificantinterpretiveissues.

Motif Assemblage Differences across the Basin

Before analysing the CA results, basic assemblage details were investigated for each of the languageareas.ThedivisionofthesitesintolanguageareawasbasedonCapell’s(1970)definedboundaries(seeFigure3.1).

figure 11.6: Bivariate plot of sample size and number of motifs recorded per sample area. The five language areas and two randomly generated Guringai samples.

language Group

code no. of sites

Darkingung 1 137Guringai 2 434

Sydney (Eora) 3 32Dharug 4 49

Tharawal 5 52

Table 11.3: language areas, codes and sample sizes.

terra australis 27

259


Darkingung

This area had 2,127 motifs (1,803 recognisable) from 137 sites. Two of the region’s four largest sites (>100 motifs) are found in this area, and average site size is 15.5 motifs/site. The motif focus here is on tracks (bird, human and roo’s respectively). The macropod is the next most common motif. Whale motifs are not represented in this assemblage. Men are the most commonly depicted human figures, followed by non-genderedanthropomorphs.Profiledepictionsarequitecommon.Relativelyfew(sixonly)cultureheroesarefoundhere.Boomerangsarethemostcommonlydepictedmaterialcultureitems(Figure11.7).

Darug

This inland area has less of a focus on tracks than its more northerly counterpart. Mundoes and birdtracksstillfigurestrongly,butmacropodsandlandanimalsrepresentalargecomponentofthisassemblage (Figure11.8).Humanfiguresareagain focusedonmales,andhere there isagreateremphasisonprofilefiguresthannon-genderedanthropomorphs.Threecultureheroes(attwo sites) are located in this area. The average site size here is eight motifs/site.

Guringai

This sample is located on the coast and represents the largest sample in the region. The 434 sites in this area produced 4,699 motifs. The average site size is 11 motifs/site. The other two sites with >100 motifs are located in this language area.

To examine the effect of sample size, two random sub-samples were generated (one with 50; theotherwith100sites)toseewhateffectthismayhaveontheresults(Figure11.9).Allthreehistograms reveal the same focus on mundoesandfish,followedbymacropods,otherlandanimalsand men. While the peaks and troughs of these graphs vary slightly according to sample size, the results are basically the same. Only the sample of 50 resulted in a reduction of motif variables (roo tracks and contact motifs). These were two of the least common motif types in the total sample. While this sample is located on the coast, its motif focus is not entirely explicable in terms of environment (cf. the Sydney group).

Sydney (Eora)

This language group is located on the coast, south of Port Jackson and north of the Georges River. This has the smallest sample size (due partly to the focus of European settlement, but also because the Cumberland Plain comprises a large proportion of this area). The 32 sites in this area produced 245 motifs (an average of eight motifs/site). The motif focusinthisareaisonfish(46%),withwhalesandothermarineanimalsalsocommon(Figure11.10).Thereducedmotifclassificationherepossiblyisaresultofsamplesize.

Tharawal

This area is also located on the coastal strip, but south of the Georges River. A sample here again was small with 245 motifs recorded at 51 sites. The assemblage sizes here (on average) are the smallest recorded in the Basin (5 motifs/site). There is a focus on marine depictions (30.7%), but the most commonly depicted motifs are men and non-gendered anthropomorphs (20.3% in total). Macropods and land animals are also common, while mundoes are the most frequently depicted tracks(Figure11.11).

terra australis 27

260


figure 11.7: darkingung language Area. motif Assemblage.

figure 11.8: darug language Area. motif Assemblage.

terra australis 27

261


figure 11.9: Guringai language Area. motif assemblage.

terra australis 27

262


Summary

Site size and motif focus vary across the region. While some of these trends are environmental; i.e.morefishandmarinedepictionsonthecoastandhigherproportionsoflandanimalsinland,there are other foci which cannot be explained so simply. There are varying proportions of animal andhumantracksindifferentpartsoftheregion,andadominanceofhumanfiguresincoastalsites south of the Georges River: The Darkingung sites are generally large with several very large sites found along access routes. The density of sites appears lower than in other areas although average assemblage size is higher. The motif focus is on kangaroo, bird and human tracks and macropods. The average Darug site size is roughly half that found in the Darkingung language area. Macropods and other land animals dominate this assemblage, but tracks appear less important. The Guringai area has the largest number of sites and seems to represent the region’s core engraving assemblage. Average site size is larger than Darug but smaller than Darkingung. Dominant motifs are mundoesandfish,followedbymacropods,otherlandanimals,andmen. Only a few sites have been recorded from the Eora area, and these are relatively small. The motiffocushereisonfish, mundoes, whales and other marine animals. The Tharawal engraving assemblages are the smallest on average. The focus here is on men,othermarinedepictions,andfish,followedbymacropodsandotherlandanimals.

figure 11.10: Sydney (Eora) Area: Motif Assemblage.

figure 11.11: Tharawal Area: motif Assemblage.

terra australis 27

263


correspondence Analysis and language AreasWhile trends in the motif assemblages across the Basin are quite clear, the CA results help determine compositional differences and foci in the different areas, and to demonstrate internal variability in the regional assemblage. The language areas were analysed as were drainage basins. This work also tested several of Capell’s boundaries along creeklines.

Drainage Basins

Atotalof25drainagebasinsweredefinedacrosstheregion(Figure11.12;Table11.4).Thesevaryconsiderablyinsizeandverylargedrainageareashavebeendefinedforareaswithlowsitenumbers (e.g. Blue Mountains and Colo). Not all basins have engraving sites; some contain only shelter art sites (chapter 12). Because of the disparate sample sizes, language boundaries and drainage basin boundaries were investigated in three locations with good sample sizes of both site types. These analyses tested the possibility of language boundaries and explored the degree of intra-language area patterning. The areas tested were:

1) Drainage basins 1, 5 and 6 within the Darkingung language area, north of the Hawkesbury River (107 sites);

2) Drainage basins 10-13 to test east-west patterning across the purported Guringai/Darug language boundary, south of the Hawkesbury River (316 sites); and,

3) Drainage basins 18 – 21 to test east-west patterning across the purported Tharawal/

Darug language boundary, south of the Georges River (51 sites).

Inthefirsttwooftheseareas,extensiverockartrecordingworkhasbeencompleted(Gunn1979;McCarthy (see references); McDonald 1986a, 1987, 1988a, 1990a; Sim 1963a, b, 1966a, b; Smith 1983; Vinnicombe 1980) as were the four excavations completed for this research. The third area hasbeenstudiedinmoredetailbyotherresearchers(Officer1984,Sefton1988,SPG1974)andalthough a smaller sample, this was seen as a useful test area south of the Georges River style boundary.

CA according to drainage basins

This analysis viewed sites according to language areas and drainage basins36(Figure11.12).Theseven clumped motif classes were again used here (i.e. ‘anthropomorphs’ include men, women, non-genderedandprofileanthropomorphicfiguresandcultureheroes).Detailonthematicfocusresults from re-inspection of the site recordings.

1) Darkingung language group (drainage basins 1, 5 and 6).

This group of 107 sites is located north of the Hawkesbury River and includes the major drainage basins of the Macdonald River and Mangrove Creek. The Upper Macdonald and Central Macdonald groups were divided at the Bala Range, a geographic barrier at the centre of the valley. Drainage Basin 4 was excluded here as it contained only one site (although now see Taçon et al. 2005, 2006).

36All plots are based on the same CA results which are sorted according to location (language area, drainage basin, etc.). The plots have a manually drawn circle or ellipse (depending on the scale of the axes) indicating the arbitrary cut-off for the homogenous ‘zone’. This procedure provides a visual aid in the interpretation of the bivariate plots. NB. Each dot symbol may represent one or many sites.

terra australis 27

264


Table 11.4: drainage Basins, language Areas and Sample sizes. Drainage Basin Basin Code Language Group No. of sites

Upper Macdonald 1 1 12Wollombi 2 1 3Wyong 3 1/2 3Colo 4 1 1Central Macdonald 5 1 55Mangrove Creek 6 1 40Mooney Mooney 7 1/2 59Brisbane Water 8 2 88Kurrajong 9 1/4 -Cattai 10 1/4 6Berowra 11 2/4 42/28Cowan 12 2 113Pittwater 13 2 27Middle Harbour 14 2 107Lane Cove 15 2/4 9/2Port Jackson 16 3 22Botany Bay 17 3 1/-Port Hacking 18 5 35Woronora 19 5 11Mill/Williams 20 4/5 -/1Georges 21 4 1/3Nepean 22 4 2Burragorang/Blue Mountains 23 4* 9Cataract 24 4?/5 -Avon/Cordeaux 25 4?/5 -

*May be mixture of Darug and Gandangara Language areas

DarkinGunG Outliers Core: 45% 6.8% 42.4%

5.1% 45.8%

The 107 sites in this language grouping are relatively homogeneous, with a heavy emphasis in positive quadrants B + C (on tracks, anthropomorphs, terrestrial animals and birds).

upper macdonald Outliers

Core: 42% 0% 14%

0% 86%

This group of 12 sites is relatively homogenous with a strong emphasis on tracks (quadrant C). Many sites here include combinations of tracks, anthropomorphs, other material objects and birds. There are no marine depictions in these sites.

terra australis 27

265


central macdonald Outliers

Core: 49% 7.1% 32%

3.6% 57%

This group of 55 sites is the most homogenous. It includes the sites along the Boree Track (a knownaccessroute).Thereisagainastrongemphasisontracksandanthropomorphicfigures(thepositive quadrants), and a number of sites which have birds only. Material objects (spears, clubs) alsofigurestrongly.Thereareveryfewmarinedepictions.Atonesite(#172)thereareplainfish,eels and turtles.

Figure 11.12: The 25 drainage basins defined across the Sydney region (refer Table 11.4).

terra australis 27

266


mangrove creek Outliers

Core: 37.5% 8% 60%

8% 24%

This group of 40 sites are the least homogenous in this language area. Outlier sites here are mostly in quadrant B, with a strong focus on terrestrial animals and material objects. Anthropomorphs andbirdsalsofigurestrongly.

2) the Guringai/Darug language boundary (drainage basins 10 - 13)

ThepurportedboundarybetweenthesetwolanguageareasisBerowraCreek(Figure3.1).Bothbanks of this estuarine waterway were surveyed during the Rock Art Project (McDonald 1990b). Vertical engravings were located on both banks of this creek, and the art on both sides (within 40m distance and 10m elevation) was observed to be very similar. This analysis tested the Berowra Creek boundary and found that the sites on either site of the creek do demonstrate differences.

figure 11.13: Bivari-ate plots, cA results. darkingung lan-guage area - upper m a c d o n a l d , c e n t r a l macdonald and man-grove creek drainage basins.

terra australis 27

267


figure 11.13: Bivari-ate plots, cA results. darkingung lan-guage area - upper m a c d o n a l d , c e n t r a l macdonald and man-grove creek drainage basins.

DaruG Outliers

Core: 35% 18.2% 45.5%

13.6% 22.7%

GurinGai Outliers

Core: 45% 27% 19% 16% 38%

The 34 Darug sites are less homogeneous than the 182 Guringai sites. There is also a general change in focus between the more unusual sites in the two areas (cf. quadrants B + C). More land animals are found in the Darug sites and more tracks, birds and marine compositions occur in the Guringai sites. The drainage basins reveal considerable variability within these groups.

cattai Outliers

Core:50% 33.3% 33.3%

0% 33.3%

Thisareahasonlysixsites,andthustheresultsaretreatedtentatively(Figure11.14).Halfofthese sites fall in the core zone and there is one each in three of the quadrants. The outlier sites havefishandeels,landanimalsandanthropomorphsandmaterialobjects.

(Darug) Berowra Outliers

Core: 32% 15.8% 47.4%

15.8% 21%

This group of 28 sites are on the left bank of Berowra Creek. These sites are quite heterogeneous with themotif focus on terrestrial animals (quadrant B: Figure 11.15).There aremany siteswith single macropods. Anthropomorphs and material objects (particularly shields) are a common combination.Cultureheroesandprofileanthropomorphsarealsopresent.Verticalengravingsarecommon.

terra australis 27

268


figure 11.14: cattai drainage Basin. Bivariate plot of cA results.

(Guringai) Berowra Outliers Core: 46% 21.7% 13% 17.4% 47.8%

This group of 42 sites is more homogeneous than those found on the western side of this drainage basin. The outlier sites here are different to those on the left bank of Berowra Creek (quadrant C: Figure11.15),withafocusonanthropomorphs,mundoes,shields,cultureheroesandotherbirds.There are more marine depictions (including whales) here. Vertical engravings are common.

Figure 11.15: Berowra Drainage Basin (Darug and Guringai Language Areas). Bivariate plot of cA results.

terra australis 27

269


figure 11.16: cowan drainage Basin. Bivariate plot of cA results.

cowan Outliers

Core: 42% 21% 22.4%

13.4% 40.3%

This group of 113 sites is quite homogeneous (Figure 11.16). Themes include a variety ofanthropomorphs, birds and other material objects (particularly axes). There is a more even spread betweenquadrantsA+B,withfishandwhalesascommonaskangaroos,shieldsandboomerangcombinations.

pittwater Outliers Core: 55.5% 66.7% 8.3% 25% 0%

Thisgroupof27sitesisthemosthomogenousofallthoseanalysed(Figure11.17).Thefocushereisheavilyonmarineanimals(quadrantA),andmanysiteshavefishonly.Therearealsomany whales.

figure 11.17: pittwater drainage Basin. Bivariate plot of cA scores.

terra australis 27

270


3) The Tharawal language area (Basins 18 – 21)

This group of 50 sites south of the Georges River fall within the Tharawal language area. There are only a few sites from the western part of this area. The 11 Woronora sites represent a slightly more ‘inland’ focus, although this creekline is only 10km from the coast and its lower reaches are estuarine.

Tharawal Outliers Core: 20% 44% 30%

10% 16%

This group of 50 sites is the least homogenous of those analysed in the region.

mill creek/Georges river: Outliers Core: 75% 0% 0%

0% 100%

The four sites in this drainage basin grouping represent too small a sample for these results to be meaningfully discussed. These results are included in the larger language group discussion.

woronora: Outliers Core: 9% 0% 70%

0% 30%

Samplesizemayalsoaffecttheresultsofthisgroup(Figure11.18).The11sitesherearehighlyheterogeneous. There is a strong focus (in quadrant B) on kangaroos and other terrestrial animals and material objects. There are less anthropomorphs and tracks (human, roo and bird). There are no marine depictions.

port hacking: Outliers Core: 17% 58.6% 10.3%

17.2% 13.8%

This group of 35 sites is also highly heterogeneous and has a different outlier focus (in quadrant A) from the preceding Tharawal group. The main compositional focus here is with marine depictions (including whales) and material objects are also common. Most assemblages in this grouparefairlysmallasisthemotifrange.Thelargestassemblagehas15identifiablemotifs;thegreatest variety of motifs at any one site is six.

terra australis 27

271


figure 11.18: Tharawal Language Area (Woronora and Port Hacking Drainage Basins). Bivariate plot of cA Scores.

Conclusions

The Darkingung and Guringai language areas have the most homogenous engraving assemblages, followed by the engraved Darug assemblage. The engravings in the Tharawal language area are the least homogenous. Internal variations across the Darkingung language area are relatively small. The upper and central Macdonald River sites are similar in terms of homogeneity and outlier focus (tracks). These sites are, however, different from the Mangrove Creek sites: which are slightly less homogenous and have a focus on terrestrial animals. The Darug sites are more heterogeneous than either the Darkingung or the Guringai sites. Capell’s language boundary along Berowra Creek is supported by this analysis. There are marked differences in the levels of variability, and dissimilar motif preferences. The Guringai area is relatively homogenous, but there is still evidence for localised variability here. The Cowan sites are less homogenous than those from the Pittwater sample. The drainage basin analysis in this central part of the region demonstrates a clinal increase in variability as one moves west away from the coast. The Mangrove Creek sites appear to be most like the Darug sites. Both groups are less homogenous than the Darkingung and Guringai sites generally. All three sets of sites, however, have different motif foci. The Tharawal sites are markedly dissimilar to all other language groups. The sites are considerably more heterogeneous and the outlier foci are different. Comparison of two drainage basins with reasonable samples within this language area suggests differences between coastal and more inland sub-groups. While having different outlier foci, these two groups demonstrate the most variance of all groups analysed. These analyses demonstrate a complex network of stylistic variability as defined byengraved motif preference across the region. Several contact language boundaries are supported

terra australis 27

272


by these analyses: the east-west Guringai-Darug boundary and the northern Tharawal boundary. Sites within the Darkingung language area, however, show some variability: with sites from Mangrove Creek more like those from the Darug language area. This patterning is discussed after compositional features and the distribution of rare motifs are described.

ridge top versus vertical engraving sitesThisresearchhasidentifiedanumberofverticalengravingsitesaroundtheforeshoresofBrokenBayanditsmainestuarinetributaries(Figure11.19).Theethnohistoricliteratureindicatesthatthis zone as a highly public one, and the sort of location where stylistic bounding behaviour is likely to be demonstrated (Wiessner 1990). On the ridgelines and plateaux above these waterways are vast numbers of open engraving sites. The art sites at the bottom of cliffs and steep hillslopes provide a different social context from the open engraving sites on the less (economically) productive plateaux above. The sites close to the estuarine resources would have been accessed either by canoe or on foot around the foreshore(Figure11.20,Figure11.21).

Many of the ridgelines around the region are documented access routes (e.g. the Boree Track, Kulnura Ridge). Sites in these locations will provide a different kind of information, one which promotessocialcohesion.Anysiteswhichhadritualsignificancearealso likely tohavebeenlocated away from the main centres of subsistence economy in any particular social group’s territory. Analysis was undertaken in the Guringai language area (Cowan and Berowra drainage basins), comparing the ridgetop, hillside and valley bottom engraving sites. This area was selected because of the large sample of engraving sites generally, and because many vertical engraving siteshavebeenfoundheretoo.Motifswerecompared(Figure11.22),andtheCAresultswerere-sorted to determine the varying degrees of homogeneity of these locations. Hillside sites were included as these locations include a high number of the engraving sites. There are 78 sites in ridgetop locations in this area. The average site size is 12 motifs/site. The average distance to permanent drinking water from these sites is 570m. The predominant motif in these locations is the mundoe (c.35%), followed by fish (15%) and men (8%).

figure 11.19: Two examples of vertical engraving sites from Berowra creek (top) and Cowan Creek (bottom). Note the complexity of composition and shared line designs.

terra australis 27

273


Figure 11.20: Smith’s Creek, Ku-Ring-Gai Chase National Park. The red stencils (arrowed and inset) in this shelter must have been produced by artists standing in a canoe at high tide. photo taken at low tide.

Figure 11.21: Shelter on Cowan Creek, the floor of which is in the littoral zone. An engraved outlined fish (inset) is located on the interior floor surface (arrowed).

terra australis 27

274


Inthe63hillslopesitesthepredominantmotifisthefish(15%)followedbymundoes (12.5%), other land animals (9%), men and boomerangs (c.8%), macropods and shields (c.7%). The average distance to permanent drinking water from these sites is 460m. There are 14 vertical engraving sites and these are generally smaller assemblages (av. 7.5 motifs/site). The average distance to drinking water from these sites is greater than from either of the other locations, for, while these are located next to the water’s edge, Berowra and CowanCreeksaresalineandtidal.Fishmotifspredominate(35%)intheselocations,followedby macropods (15%). Missing motif classes include culture heroes, circles and contact motifs. Mundoes are present in a very small number of these sites (c.2%). The differences in motif preferences on ridgetop and hill side locations are quite striking andcannotbeexplainedintermsofsampling(Figure11.22).TheCAresultsreinforceengravingsitedifferencesintheselandscapes(Figure11.23).

ridgetops Outliers

Core: 45% 26.2% 21.4%

9.5% 42.9%

figure 11.22: motif histogram for engraving sites in ridgetop, hillslope and vertical engravings in estuarine valley bottom.

terra australis 27

275


This group is quite homogeneous, with its outlier focus in quadrant C. There is a slight subsidiary focus in quadrant A.

hillslopes Outliers

Core: 48% 18.8% 18.8%

18.8% 43.8%

This group of 62 sites has a similar level of homogeneity to the ridgetop sites, and the same outlier focus (quadrant C). The equal distribution between the three other quadrants indicates a broader range of subject combinations in these sites.

vertical sites, valley bottoms Outliers

Core: 36% 33.3% 11.1%

0% 55.6%

This group of 14 sites is the most heterogeneous, but has the same outlier focus as the other two groups. The sites on the ridgetops and hillslopes are more homogenous than those around the water’s edge. While there are demonstrated differences in degrees of homogeneity, however, the outlier foci in these three locations are very similar. In other words, the graphic vocabulary of people operating in these different landscapes, is the same. This result is as would be expected within the one language area and suggests that these sites being used by the same group(s) of people in a range of different social or information contexts.

rare motifsRare and unique motifs were analysed to establish their geographic distributions. It was hoped that this type of investigation would elucidate localised stylistic traits. Analysis concentrated on non-economicmotifsinanefforttoreduceenvironmentalinfluences. The analysis of how many times an individual motif occurred at any engraving site in the regiondemonstratedsomeinterestingresults(Table11.5;Figure11.24). Themotifswhichoccurat themostsites in the regionarefishandmacropods (c.35%)followed by men and other land animals. Mundoes – while the most numerous engraved motif in the region - are only present at 22% of engraving sites. This result indicates that certain motifs are concentrated in a few sites - while other motifs are more widely dispersed (relatively fewer motifs are placed on many more sites). This analysis focussed on concentrated rare motifs (i.e. ones which are relatively rare and which occur on few sites), and on dispersed motifs (i.e. relatively rare but fairly widely distributed). Thedistributionofthesemotifswasplotted(Figure11.25toFigure11.28),thepercentagesof sites with each motif type were calculated by language area (Table 11.6). These results were comparedwiththepercentageresultsforeachlanguageareasothatrelativesignificancecouldbedetermined.Totestthestatisticalsignificanceofthesedifferences,anapproximaterandomisationmethod(Noreen1989,Wright1991)wasusedonthefigures(seeTable11.7).

terra australis 27

276


Table 11.5: engraving sites. motif total, maximum incidence at any particular site, number of sites in the region with motif present, and % of sites with motif.

motif Total max incidence Sites with motif present

% of Sites with Motif

Man 422 14 199 27.8Woman 79 5 56 7.8Anthropomorph 182 15 115 16.1ProfilePerson 79 5 53 7.4Culture Hero 36 2 29 4.1Macropod 543 13 247 34.5Snake 56 3 45 6.3Other Land Animal 312 10 170 23.7Emu 76 7 51 4.1Other Bird 166 11 101 14.1Fish 905 47 250 34.9Eel 182 7 97 13.5Whale 101 7 71 9.9Other Marine 156 9 97 13.5Shield 232 23 103 14.4Boomerang 303 15 144 20.1Axe 45 5 28 3.9Other material object 218 13 106 14.8Mundoe 1,360 99 157 21.9Roo track 186 18 43 6.0Bird track 541 95 71 9.9Circle 309 17 96 13.4CXNF 70 3 53 7.4contact 36 11 11 1.5hand 19 3 11 1.5

figure 11.23: cA bivariate plots according to topographic location. Guringai language area: Berowra and cowan drainage basins.

terra australis 27

277


figure 11.24: engraving component. percentage of sites at which particular motifs appear.

Thisanalysisconfirmedsomeofthedisparitiesrevealedbythepreviousanalyses:

Thereisasignificantly higherproportionofprofilefiguresandrootracksandasignificantlylower proportion of shields and axes in the Darkingung area.

Shieldsandaxesoccurinsignificantly higher proportions in the Guringai area, while there aresignificantlyfewerprofilepeopleandcomplex-non-figurativemotifsinthisarea.

In the Darugarea,therearesignificantlyhighernumbersofprofilepeopleandcomplex-non-figurativemotifs,andcontactmotifsandhandsoccuroften.

In the Tharawalarea,therearesignificantlylowerproportionsofwomen,snakes,shieldsand axes.

Significantdifferencesareidentifiedbetweenlanguageareas(Table11.7).Thedifferencesand similarities between the Guringai and Darkingunggroupsarestatisticallysignificant.The differences (and similarities) between the Guringai sites and the Darug and Tharawal sitesarealsosignificant.

Table 11.6: Rare Engraving Motifs. Distribution per Language Area (outstanding results in red and bold).

Motif Number (and %) of Sites with motif in each Language AreaDarkingung Guringai Eora Darug Tharawal

Woman 12 21.8 40 72.7 0 0 2 3.6 1 1.8Profile person 26 49.1 22 41.5 0 0 5 9.4 0 0Culture hero 5 17.2 22 75.9 0 0 2 6.9 0 0Snake 11 24.4 27 60.0 1 2.2 4 8.9 2 4.4Shield 11 10.7 88 85.4 0 0 3 2.9 1 1.0Axe 3 10.7 23 82.1 0 0 1 3.6 1 3.6Roo tracks 30 69.8 7 16.3 0 0 4 9.3 2 4.7CXNF 17 32.1 24 45.3 2 3.8 6 11.3 4 7.5Contact 3 18.2 6 72.7 0 0 2 9.1 0 0Hand 2 18.2 8 72.7 0 0 1 9.1 0 0

Total sample 2127 (19.5) 4699 (61.9) 245 (6.5) 360 (4.6) 245 (7.4)

terra australis 27

278


Theseresultsconfirmthelocalisedcharacterof the engraving assemblage in different areas of the Basin. The distributions of the rarer motifs also demonstrate some interesting connections. Sites with culture heroes occur mostly in the western part of the Guringai territory and into the Darkingung territory. This

design link is not suggested by the CA results which show the overall foci to be quite different from thesetwoareas.Profileanthropomorphs,axesandcontactmotifshaveverysimilardistributionsto the culture heroes. This will be discussed further below.

compositionDifference in composition were explored to provide further evidence of the types of cultural and/or stylistic choices (Sackett 1990) being made across the region. Shields and culture heroes were selected for this analysis.

Shields

Previous archaeological analyses of shield designs have demonstrated stylistic patterning explicable in terms of trade and overall alliance systems (Dickens 1992, Hatte 1992; Morwood 1987). Local ethnohistoric evidence suggested that this motif type may provide ethnically-sensitive patterning, with commentators stating that the coastal peoples from around Sydney and further north carried distinctively patterned shields (Bellinghausen and Rossiyisky in Barratt 1981; Enright 1900; Threlkeld in Gunson 1974). It was hoped that patterning in shield design distributions may indicate an interrelatedness of contact around the region. Threlkeld described the construction of the region’s wooden shields from around the Lake Macquarie area (Awabakal/Guringai language areas). These were:

three feet long by eighteen inches ... lozenge shaped, pointed at top and bottom, and pigeon breasted rather than flat. ... The shields are always painted with white pipeclay and aregenerally ornamented with a St George’s Cross, formed by two bands two or three inches wide, one vertical the other horizontal, coloured red ... [Gunson 1974:68].

Rossiyisky describes the wooden shields from the Sydney area similarly, although observing that ‘they are daubed with variousredandwhitefigures’[(in)Barratt1981:23,emphasismine].Bellinghausen’sdescriptionconfirms the colourusageas ‘drywhite colouring substanceoverwhich was painted red stripes’ (in Barratt 1981:41; and see Enright 1900; Cave in Brayshaw 1986). The engraved shields are fairly rare (232 total: 3% of the assemblage) and they are quite dispersed (found at 103 sites; 14%). A few sites have concentrations of this motif type: site# 45-3-376has23shields;#45-6-705has15(Figure11.29).Mostsiteshaveoneortwoshields(average is 2 shields/site). These motifs are usually found on moderately large sites (average assemblage size = 22 motifs/site). The shield from one site (#45-6-689) was excluded as the motif was incomplete. The sample for this analysis came from 102 sites.

Language Areas compared Significance value

Darkingung Guringai <.001Guringai Darug .013Guringai Tharawal .027

Table 11.7: Engraving sites. Significant values for rare motifs in the five language areas.

terra australis 27

279


Figure 11.25: Distribution of sites with engraved women and profile people.

terra australis 27

280


figure 11.26: distribution of engraving sites with culture heroes and shield motifs.

terra australis 27

281


figure 11.27: distribution of sites with engraved axe and macropod tracks.

terra australis 27

282


Figure 11.28: Distribution of engraving sites with complex-non-figurative and contact motifs.

terra australis 27

283


Figure 11.29: Site 575 (NPWS # 45-6-705) with 15 shield motifs and three boomerangs.

Most shields (85.4%) are in the Guringai language area. Darkingung has the next highest number (10.7%) followed by Darug (2.9%) and Tharawal (1.0%). No engraved shields are recorded in the Eora language area. Given the paucity of design options provided by ethnohistoric accounts, the engraved shield assemblage was inspected to determine design variety. Analysis revealed this to be considerably greater than suggested by Threlkeld’s description. Twenty-six design categories (including undecorated)wereidentified(Figure11.30).Thedesignsconsistmainlyofhorizontalandverticallinevariations.Asignificantdesignelementisadiamondshapedcomponentateitherorbothofthe shield’s pointed end(s). There are several unique varieties (designs 6 A-C) which are variations on design themes 2B and 2C. Two of these appear to be shields punctured by many spears (see Megaw 1993). The two sites with multiple shield motifs (#’s 45-3-376, 45-6-705) are separated by considerable distance. One is north of the Hawkesbury River (Mangrove Creek drainage basin), theother is atMosman (MiddleHarbourdrainagebasin:Figure11.31).TheDarkingung site includes only design varieties 1, 2B and 2C. The Guringai site has seven design types (4A, 4D, 4E,4F,5B,5E,and2C)mostofwhichhavethediamondpointdecorativeelementatoneorbothends. First, siteswithonlyasingledesign typepresentwereanalysed. Itwashoped that thiswould facilitate clear focus on localised patterning in the design elements and reduce synchronic ‘noise’. Threlkeld’s description suggested that the ‘St George cross’ form (2B) would be the most common. This was not the case. Sites with multiple design types were excluded from this initial sort.Fourmajordesignthemeswereidentifiedandtheseaccountfor62%ofthesiteswithshields(Table 11.8).

Table 11.8: engraved sites. Shield design Types according to language areas and drainage basins. Sites with single design types only.

Language Area and Drainage Basins

Plain % 2B % 2C % 4+5 variation

% %f

Darkingung cent Macdonald 1 6.7 1 11.1 2 7.1 6.3Darkingung Mooney 3 20.0 4.8Darug Berowra 1 6.7 1 3.6 3.2Darug Lane Cove 1 3.6 1.6Guringai Mooney 1 6.7 4 14.3 7.9Guringai Brisbane Water 2 13.3 5 17.9 11.1Guringai Cowan 3 20.0 3 33.3 7 25.0 7 63.6 31.8Guringai Pittwater 1 11.1 2 7.1 4.8Guringai Middle Harbour 3 20.0 4 44.4 5 17.9 4 36.4 25.4Guringai Lane Cove 1 3.6 1.6Tharawal Port Hacking 1 6.7 1.6Totals 15 (23.8) 9 (14.3) 28 (44.4) 11 (17.4) 63

terra australis 27

284


Plain shields, as might be expected, are ubiquitous. They occur on both sides of the Hawkesbury River in Guringai, Darkingung, Darug and Tharawal sites. The distribution of the St George cross design (2B) was quite restricted. With the exception of one Darkingung site west of Mangrove Creek(onFlatRocksRidge),thisdesignisconfinedtotheCowan,Pittwaterand(particularly)Middle Harbour drainage basins: all Guringai language areas. The double (horizontal) cross design (2C) is the most common design in the region. It occurs at 28 sites (in isolation) and at another 13 sites in combination with a variety of design forms. It occurs in Guringai, Darkingung and Darug areas. The design with diamond elements at either end of the shield has a very restricted distribution. The eleven sites with this design element (in isolation) occur only in the Guringai area, in the Cowan and Middle Harbour drainage basins. The two designs which appear to have the most potential for indicating inter- and intra-language contact were the St George (2B) variety and the diamond-end varieties (design types 4 and 5), both of which are restricted predominantly to the Guringai area. Next the distribution of these design elements in sites with combination shield designs was analysed. Again, a restricted distribution was found (Table 11.9). More than 90% of the sites with these designs were in the Guringai area, with the Cowan and Middle Harbour catchments containing 68% of these sites. Shield designs exhibit highly localised characteristics. The Guringai language area contains the largest number of shields in the region and exhibits the greatest degree of design variability. There are localised design traits within this area. Cowan and Middle Harbour catchments containthemostshieldsandthisisthedesignfocusforboththediamondinfillanddouble-crossvarieties.

figure 11.30: range of shield designs present in the engraved component.

terra australis 27

285


figure 11.31: distribution map of main shield designs.

In this instance, the ethnohistoric literature is highly inadequate with respect to both the design variability present and the distribution of these design elements. There may have been a major constriction in design elements used on shields at contact (and the St George Cross may have predominated as reported). Conversely, the subtleties of design variability were overlooked by early observers. If the former explanation is correct, then the ethnohistoric ‘present’ has little applicability in the prehistoric past, particularlyintermsofethnicallysignificantdesigns. This would appear to be another

Area No. %Darkingung central Macdonald 1 4.5Darug Berowra 1 4.5Guringai Mooney 1 4.5Guringai Brisbane Waters 3 13.3Guringai Cowan 7 31.8Guringai Pittwater 1 4.5Guringai Middle Harbour 8 36.4

Totals 22

Table 11.9: Sites with engraved shield. Shields distribution of mixed design types according to language areas and drainage basins. Type 2c, 4 and 5 varieties.

terra australis 27

286


example of the inadequacies of European observation regarding the richness of Aboriginal culture in the region.

Culture Heroes

At 29 sites across the region there are 36 culture heroes (Table 11.10). This motif differs from ordinary anthropomorphs on the basis of extreme size (mean length is 5.2m; standard deviation =1.4m)andbytheamountofinfilleddecoration.Someofthesemotifsaretherianthropes(Flood1987; McDonald and Veth 2006b) with animal features including bird’s and/or snake’s heads. The two main forms of culture hero are the ‘Daramulan’type(partiallyorfullyinprofile)andthe ‘Biaime’ type (in plan, with limbs akimbo). These types were named by McCarthy (1959a; following Mathews 1904, and see Clegg 1981) and a more recent analysis (Higgs 2003) has explored attribute variability in these motif forms. A feature of this motif form suggests its cultural importance - or at least its continuation of use over time: evidence for multiple episode of engraving. More than half of these motifs have added features or altered outlines. The Daramulanfiguresappearparticularlysusceptibletochange:manyhavingasecondlegadded,andsometimesanarmorextrapenis(Figure11.33). These motifs occur amongst a range of assemblage sizes. Some sites are extremely large (97 motifs), but there are also isolated examples e.g. with only a single culture hero present. The average size of sites with this motif present is 25.4 motifs/site. The average distance to drinking water from sites with these motifs is c.500m. As indicated above, this motif is restricted to the Guringai (75%), Darug (8.3%) and Darkingung (16.7%) language areas. Almost half of the motifs in the Guringai area are located in the Cowan drainage basin (Table 11.11). Most of these motifs occur singly at sites, but are often paired. Some of these pairings are clearly male/female; others are male/male. Some sites include paired and/or transitional Daramulan and Biaimetypes(Figure11.34). The Daramulan type is heavily focussed (86%) in the Guringai area, while the Biaime form is distributed more widely (still 67% Guringai). The two Daramulan-type culture heroes located outside the Guringaiareaoccuronmajorengravingsites(Devil’sRockMaroota,FlatRocks Ridge) in places which could be argued are aggregation locales (Conkey 1980) - or at least on access routes where groups cohesion is being demonstrated.

Table 11.10: engraving sites. culture heroes: compositional details.Site

NumberMax

length (m)

Type Biaime/

Daramulan

Animal features

Infill Other features

37-6-8 2.7 B dot breasts, penis and foot added, fingers 1 hand, 2 eyes, headdress, 47 dots

45-2-16 6.8 D bird head heavy dot/linear

penis + leg added, ornate headdress, bird head

5.0 B - heavy dot/linear

pointy ears headdress, male, dotted linear infill

45-2-45 3.3 B - heavy two deep eyes, body linear infill waist band no hands or feet, conical head

3.0 B - some lines

two deep eyes + 3 smaller ones, 4 dots in penis; toes on feet, stumpy arms

45-3-39 4.5 D birdlike linear bird like with long rayed headdress. No foot visible. Waistband + linear infill

45-3-56 5.0 D birdlike linear bird-like with long pointed headdress, 2 eyes + mouth. No foot visible

4.5 D birdlike linear bird like with rayed headdress, 4 eyes Waistband + 2 rows infill

45-3-99 8.3 B - lines dots twisted perspective; penis to one side45-3-110 3.2 D birdlike linear

dotsleg with human foot, head twisted perspective w headdress. 2 eyes

terra australis 27

287


Site Number

Max length

(m)

Type Biaime/

Daramulan

Animal features

Infill Other features

45-3-168 5.7 D bird head lines on neck

emu like; 2 legs w feet and penis to side

45-3-228 6.0 B long ears lines buttocks shown + 2 legs, enormous ears, breasts? under arms, + 5 eyes

45-3-232 5.6 B bird head lines no arms, twisted perspective head, headdress

45-3-954 3.5 B - heavy linear

feet, fingers 1 hand, no penis or headdress, 2 eyes

45-3-1289 6.5 B - lines eyes, headdress45-6-42 4.7 D - lines 2 penises, eyes, headdress45-6-44 4.6 B - lines large head, no neck; 1 leg added; toes 1

foot only; single line headdress, 3 eyes45-6-85 5.2 D snake

head- snake + other head with headdress. Human

foot45-6-284 3.4 B - linear 7 dots on face no ears; fingers + toes;

fringed body infill, big penis; holding boomerang assoc. with long trail mundoes

45-6-290 5.3 B - lines 2 eyes, no feet6.0 D snake/

bird- 1 leg, sinuous neck, arm + axe added,

profile leg6.0 B/D snake

head- 2 legs, sinuous neck, 1 arm with axe, 2 legs

with feet (1 added) penis added45-6-312 5.0 B pointy

earslinear design

fingers toes, facial features, holding a number of material objects, body design.

45-6-313 4.5 B - lines very long arms and legs foot on one leg only, fingers, toes

45-6-315 6.0 B - lines fingers + toes, waist + arm bands, small penis, looped headdress; associated with shields

45-6-316 5.3 B - many lines

horizontal + vertical infill; fingers on arms and legs, no neck, facial features, penis

4.7 B - many lines

horizontal + vertical infill finger like appendages on arms and legs, no neck, facial features include an inverted ‘smile’, penis with infill

45-6-323 5.0 B - lines hands + feet with fingers + toes, barbed arrow in side, crossed body lines, penis

45-6-324 6.3 D bird like lines emu like with human foot, arm added45-6-346 4.2 B - lines rayed headdress, girdle + arm bands, 2

eyes3.2 B - lines rayed headdress, girdle + arm bands, body

design,1 foot45-6-412 6.7 D - lines male with key shaped headdress

5.5+ D - - female, pointed breast, 5 eyes45-6-434 5.7 B - lines waist + armbands, penis, feet with toes,

headdress, holding fish 45-6-436 9.3 D - lines leg added on to penis, arms added onto

head; headdress45-6-890 6.2 D bird head lines foot with toes, arm + hand with fingers

added. Birdlike head with beak. Long meandering penis, 6 eyes, dot on heel

They focus of both culture hero forms in the Guringai area, with relatively minor examples in northerly and westerly language areas, suggests design (and social) contact between these three language areas, with design focus stemming from the Guringai. A similar pattern was demonstrated by the shields. The distribution of these same motifs in the shelter art component is compared below, to determine whether these patterns hold for both media.

terra australis 27

288


Table 11.11: daramulan and Biaime types, language areas and drainage basins.Area Daramulan % Biaime % Total %

Darkingung U McDonald 1 100 1 2.8Darkingung C McDonald 4 100 4 11.1Darkingung Mangrove 1 100 1 2.8Darug Cattai 1 100 1 2.8Darug Berowra 1 50 1 50 2 5.6Guringai Mooney 1 50 1 50 2 5.6Guringai Brisbane Waters 3 75 1 25 4 11.1Guringai Berowra 3 60 2 40 5 13.9Guringai Cowan 5 31.3 10* 62.5 16 44.4Totals 14 (38.9) 21 (58.3) 36 100.1

*one of these culture hero is half Daramulan and half Biaime.

figure 11.32: distribution of daramulan and Biaime type engraved motifs.

terra australis 27

289


figure 11.33: engraved culture heroes. daramulan motifs which appears to have been altered over time (top to bottom, L to R; after McCarthy 1956: group 3, Figure V; McCarthy 1954a: Figure 7C; Campbell 1899: Plate XXV, Fig 4; McCarthy 1956: Group 3, Figure V).

engraving sites: conclusionsWhileacohesivestyleregion,theseanalyseshaveidentifiedstylisticvariabilityintheengravedart of the Sydney region. This is distinctive only in the south of the Basin, where the previously identifiedstyleboundaryoftheGeorgesRiverisconfirmed.Inmostpartsoftheregionstylisticvariationisclinal.Itcanbeexplainedintermsofdefinedlanguageareas,andcanalsobeidentifiedas localised variability based on drainage basins. The Darkingung and Guringai language areas have the most homogenous engraving assemblages closely followed by the Darug. The engravings in the Tharawal area are the least homogenous. The Guringai sites consistently demonstrate the highest levels of homogeneity, but also provide evidence for internal variability: the Cowan sites are less homogenous than those from the Pittwater drainage basin. In this central part of the region there is a clinal increase in variability as one moves west, away from the coast.

terra australis 27

290


figure 11.34: paired and/or transitional Daramulan and Biaime motifs. note that Daramulan below (on #45-2-16) has an altered outline also (top; from McCarthy and Hansen 1960: Figure 1; bottom; from McDonald 1986a: Figures 3 and 8).

terra australis 27

291


Analysis of the Tharawal sites reveals an assemblage which dissimilar from all others. The sitesaresignificantlymoreheterogeneousandtheoutlierfociindicateadifferentthematicfocus.Comparison of the two drainage basins with good samples here suggests differences between the coastal and more inland sub-groups. The results demonstrate a complex network of stylistic variability as defined bymotifpreference. In some language areas, sites show levels of internal cohesion. Several of the proposed language boundaries are supported by these analyses. Berowra Creek could be the boundary between the Darug and Guringai groups – the art on either sides of this showing varying design focus. The northern boundary for Tharawalisalsosupportedbytheseanalyseswiththeidentifiedstyle boundary at the Georges River. Theanalysisofraremotifsconfirmsthelocalisedcharacterofengravingthemesaroundthe Basin. Culture heroes are focussed in the western part of the Guringai territory and reveal a design link with the Darkingung -confirmingtheCAresults. The study of composition on several rare motifs also revealed design contact between Guringai, Darkingung and Darug language areas. The source of this contact appears to stem from the Guringai. Design variability on shield motifs is extraordinarily diverse in the Guringai area, with both less motifs and a marked decrease in design options being practised outside this language area. These results will be discussed further in chapter 13.

Documents

SYNCHRONIC vARIATION: SYDNEY BASIN ENGRAvED ARTpress-files.anu.edu.au/downloads/press/p109931/pdf/ch111.pdf · terra australis 27 253 Chapter 11: Synchronic variation: Sydney Basin